[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$f18v95r025ucxh":3},{"code":4,"message":5,"data":6,"timestamp":28},10000,"success",{"full_name":7,"position":8,"description":9,"language":10,"stars":11,"forks":12,"stars_gained":13,"age_days":14,"hn_threads":15,"one_liner":16,"summary_md":17,"topic":18,"github_topics":21,"license":22,"homepage":23,"updated_at":24,"history":25},"tradesdontlie\u002Ftradingview-mcp",7,"AI-assisted TradingView chart analysis — connect Claude Code to your TradingView Desktop for personal workflow automation","JavaScript",6942,2853,16,210,[],"Bridges Claude Code to a locally running TradingView Desktop so an LLM can read, draw on and script your own charts.","If you already pay for TradingView and script Pine indicators, this is a practical experiment: it gives an LLM eyes and hands on the chart you are looking at, rather than on a data dump. The catch is that it depends on undocumented internals of the desktop app, so treat it as a workstation helper, not infrastructure.\n\n## The problem\n\nA discretionary chart user wants an assistant to answer questions about the exact chart in front of them: what the indicator table currently reads, where the marked levels are, how a Pine script behaves on this symbol and timeframe. Normally the user has to copy values into a chat, describe the layout, paste script errors back and forth, and re-run each edit manually. This project changes that loop by attaching an LLM to the running desktop app, so it can read chart state, navigate symbols and timeframes, edit and compile Pine scripts, draw annotations, manage alerts, and step through replay bars. The result is a conversational analysis and scripting session on the same chart, with screenshots available for visual analysis.\n\n## What's different\n\nThe mechanism is straightforward in principle: TradingView Desktop is an Electron app, so it can be launched with Chrome DevTools Protocol enabled on port 9222, and the MCP server talks only to that local instance. The README frames this as a standard Chromium debugging interface and states that no TradingView data is transmitted externally, nothing in the app is modified, and no paywall is bypassed. Claude Code consumes the tools through an MCP configuration, and the project also exposes a CLI path, so an assistant is not strictly required for every operation.\n\nThe practical benefit is that the model works against live chart state instead of a pasted snapshot, which is what makes iterative Pine development and \"what is my chart showing right now\" questions plausible. The trade-off is structural: the README says the tool accesses undocumented internal TradingView APIs and can break with any TradingView update; pinning the desktop version is the suggested mitigation. It also explicitly does not execute real trades, so it is an analysis and authoring aid, not an execution system. The README's research framing is honest about the open questions: latency, context and how the agent copes with ambiguous financial UI state are things this project explores rather than claims to have solved.\n\n## Who should try it\n\nA good fit is an active TradingView subscriber who already writes Pine Script, runs Claude Code, and is comfortable launching an app with a debug flag. For that person, the ability to have an assistant inspect indicator values and iterate on a script against the actual chart removes a lot of manual copying, and the MIT-licensed source is available for local inspection. It is a poor fit for anyone hoping to automate trading or get market data without a subscription: there is no order execution, and the tool will not work without the desktop app and a valid plan. Someone who mostly reads TradingView in a browser or on mobile has nothing to attach to. A trader who needs a stable, continuously running integration should weigh the version-pinning caveat before adopting it. Readers curious about the broader experiment can check the [MCP listing](https:\u002F\u002Fmcptoplist.com\u002Fserver\u002Fglama%2Ftradesdontlie%2Ftradingview-mcp).\n\n## Requirements and limitations\n\nYou need the TradingView Desktop app with a subscription, Node.js 18+, and macOS, Windows or Linux. Claude Code with MCP support is required for the MCP tools, but the README also documents a CLI mode usable from any terminal, so the assistant is not the only entry point. Setup involves cloning the repo, installing dependencies, and adding the server to your MCP config at `~\u002F.claude\u002F.mcp.json` or a project `.mcp.json`; the launch scripts or manual `--remote-debugging-port=9222` flag start TradingView in debug mode.\n\nOngoing upkeep is the main constraint. Because the bridge relies on internal Electron structure, a TradingView update can break it without notice, and the README recommends pinning the desktop version if stability matters. The license covers only this source code, not TradingView's software or data. The disclaimer also notes that this is an unaffiliated project, so users should read it before relying on it in a workflow.\n\n## HotRepo's take\n\nThe only recorded HotRepo observation is 2026-10-25, when the repository had 6,938 stars and appeared on the daily board at a +16 day. That is a single snapshot, so persistence cannot be judged from it; the 209-day age and the star count describe the repository, not its maintenance quality. The practical judgement is that this is a narrowly scoped experiment with a clear dependency risk, and the one thing worth verifying before committing is how tightly a pinned TradingView Desktop version pairs with the current tool set.",{"slug":19,"name":20},"finance","Finance",[],"NOASSERTION",null,"2026-10-25T11:04:50Z",[26],{"date":27,"stars":11,"stars_gained":13},"2026-10-25",1792968455]